Comments on: Getting it wrong on multicore

Macrovision's David Znidarsic says multicore processing is leading some software companies to make poor decisions about licensing.

[aabcdefghij987654321]

Charge by CPU didn't make much sense anyway

After all, you got charged the same for a four processor system using 200Mhz Pentium Pro CPUs as you would for one with 3Ghz Pentium IVs despite the huge gap in performance between those systems.

There are no easy answers but stubbornly clinging to the per/CPU model like Oracle is doing is definitely not right.

[orfeu_niko]

It is about rights

A vendor has the right to charge how much he wants and on any license scheme he chooses.
The customer has the right to choose to buy, or not, a product.
I read a lot about disactisfaction and still those people choose to buy it. I know, it is not black and white, but when you feel it is enough, you should take messures.

It Depends

Your assumption is that all software is used on the desktop, and can be priced according to some sort of "per seat" model. However, there's a large contingent of enterprise software, like middleware, can't be sold on a "per seat" basis because most users never see it or actually touch it. It lives on the network, basically running a ton of tasks connecting applications and the like. So in those instances, the only model that works is the CPU model. BTW, many of Oracle's offerings fit into this category.

[aabcdefghij987654321]

Charge by CPU didn't make much sense anyway

After all, you got charged the same for a four processor system using 200Mhz Pentium Pro CPUs as you would for one with 3Ghz Pentium IVs despite the huge gap in performance between those systems.

There are no easy answers but stubbornly clinging to the per/CPU model like Oracle is doing is definitely not right.

[orfeu_niko]

It is about rights

A vendor has the right to charge how much he wants and on any license scheme he chooses.
The customer has the right to choose to buy, or not, a product.
I read a lot about disactisfaction and still those people choose to buy it. I know, it is not black and white, but when you feel it is enough, you should take messures.

It Depends

Your assumption is that all software is used on the desktop, and can be priced according to some sort of "per seat" model. However, there's a large contingent of enterprise software, like middleware, can't be sold on a "per seat" basis because most users never see it or actually touch it. It lives on the network, basically running a ton of tasks connecting applications and the like. So in those instances, the only model that works is the CPU model. BTW, many of Oracle's offerings fit into this category.

[Al Cook]

Derived value a poor basis for licensing

Maybe I'm old fashioned, but I think that software vendors should be paid for the work they do, hardware vendors should be paid for the work they do, and smart people in IT departments should be paid for the work they do. If I purchase a more powerful server from a hardware vendor, I don't believe I should owe the software vendor more money for a license for which I already paid them, and for which they happily took my money, just because I can get more transactions per second. In effect, the software vendor is charging a tax on my hardware upgrade. In the case of many of these companies that tax is more than the entire revenue recognized buy the hardware vendor. For example, if I replace a single-proc MS SQL database server with a double-proc server, I might pay the hardware vendor $5,000 for the server, but I own Microsoft another $15,000 for another SQL Server proc license. What did MS do for that extra revenue? Exactly nothing. (And Oracle's tax is probably even higher.) Licensing for derived value is not only an enormously complicated metric (Oracle tried it and dropped it) but it is fundamentally unfair. Software vendors should determine what they think their software is worth and charge customers accordingly, regardless of the hardware that it runs on. If they want to create versions that deliver varying amounts of power, e.g., a DB limited to 2GB databases and another that will handle 1TB, fine. But don't tax hardware vendors.
Al Cook

[Al Cook]

Derived value a poor basis for licensing

Maybe I'm old fashioned, but I think that software vendors should be paid for the work they do, hardware vendors should be paid for the work they do, and smart people in IT departments should be paid for the work they do. If I purchase a more powerful server from a hardware vendor, I don't believe I should owe the software vendor more money for a license for which I already paid them, and for which they happily took my money, just because I can get more transactions per second. In effect, the software vendor is charging a tax on my hardware upgrade. In the case of many of these companies that tax is more than the entire revenue recognized buy the hardware vendor. For example, if I replace a single-proc MS SQL database server with a double-proc server, I might pay the hardware vendor $5,000 for the server, but I own Microsoft another $15,000 for another SQL Server proc license. What did MS do for that extra revenue? Exactly nothing. (And Oracle's tax is probably even higher.) Licensing for derived value is not only an enormously complicated metric (Oracle tried it and dropped it) but it is fundamentally unfair. Software vendors should determine what they think their software is worth and charge customers accordingly, regardless of the hardware that it runs on. If they want to create versions that deliver varying amounts of power, e.g., a DB limited to 2GB databases and another that will handle 1TB, fine. But don't tax hardware vendors.
Al Cook